Marco Biagi will defend his PhD thesis on June 19, 2026, at CEA/SPINTEC in Grenoble, presenting an exploration of orbital-to-spin conversion materials for three-terminal spin-orbit torque (SOT) magnetic tunnel junctions (MTJs). The work tackles persistent challenges in SOT-MRAM: limited write-current efficiency and high resistivity in standard heavy-metal-based structures. Recent research suggests orbital currents offer a path to higher torques, as orbital Hall angles can far exceed their spin counterparts and exist in a wider range of low-resistivity materials. Since orbital currents alone cannot torque magnetization, an additional conversion layer is needed to transform them into spin currents.

Biagi evaluated two families of orbital-source/conversion-layer stacks integrated with FeCoB: Ru/HM/FeCoB and Ta/W/FeCoB. Ru and Ta serve as orbital current generators, while the heavy metals (HM = Ta, W, Pt) convert orbital to spin current. Ru is predicted to have one of the largest orbital Hall angles among transition metals with low resistivity; Ta’s orbital Hall angle is estimated to be roughly an order of magnitude larger than its prominent spin Hall effect. When a conversion layer is present, multiple spin-current contributions can add linearly, potentially enhancing the total effective spin Hall conductivity.

Measurements of saturation magnetization, effective anisotropy field, resistivity, and damping-like (ξDL) and field-like (ξFL) SOT efficiencies were performed as a function of layer thickness, both as-deposited and after 300°C annealing, with benchmarking against reference HM/FeCoB stacks. For Ru/Ta and Ru/W, the enhancement in ξDL was limited. In contrast, Ru/Pt showed a twofold increase in ξDL compared to Pt alone, which is attributed to Pt’s strong spin-orbit coupling enabling efficient orbital-to-spin conversion. The independence of ξDL from Ru thickness indicates an interfacial origin. However, thermal annealing strongly degraded ξDL in Ru/Pt, limiting its practical use.

Ta/W stacks exhibited a much stronger boost: ξDL increased by a factor of 4.4 relative to Ta and 3.2 relative to W. A parallel-resistor model showed that conventional spin Hall contributions alone cannot explain this gain, pointing to an orbital mechanism. Moreover, after 400°C annealing, ξDL remained stable and perpendicular magnetic anisotropy was preserved, demonstrating excellent thermal robustness.

Taking advantage of these properties, Ta/W was integrated into full SOT-MTJs and benchmarked against standard W-based devices. Ta/W MTJs achieved comparable critical switching currents but with lower switching current density and improved perpendicular anisotropy stability—the first demonstration of integrated orbital-to-spin conversion in a working SOT-MTJ. A proof-of-concept for vertical non-local switching using orbital torques was also presented, which simplifies fabrication of bottom-pinned SOT-MRAM. Overall, the thesis validates orbital physics as a means to enhance SOT-MTJ performance and move toward scalable MRAM technologies.

On June 19th 2026 at 14:00, Marco Biagi (SPINTEC) will defend his PhD thesis entitled : Exploration of orbital-to-spin conversion materials and integration in 3-terminal spin-orbit torque magnetic tunnel junctions

Place : IRIG/SPINTEC, CEA Building 10.05, auditorium 445 (presential access to the conference room at CEA in Grenoble requires an entry authorization, request it before June 8th to admin.spintec@cea.fr)

visio conference : https://univ-grenoble-alpes-fr.zoom.us/j/98769867024?pwd=dXNnT3RMeThjYStybGVQSUN0TVdJdz09

Meeting ID: 987 6986 7024

Passcode: 025918

Abstract : The development of electrically controlled nanomagnets for spintronic applications, particularly non-volatile magnetic memories (MRAM), is attracting strong interest due to the limitations of CMOS-based memories such as SRAM and eDRAM. Spin–orbit torque (SOT) MRAMs are promising candidates for addressing SRAM specifications; however, current materials still suffer from limited efficiency and high resistivity, leading to unmet write-current requirements. Recently, studies have highlighted orbital phenomena as a potential route to enhance SOT efficiency, owing to their larger magnitudes and availability in a broader set of materials. However, orbital currents do not couple to magnetization in the absence of spin–orbit coupling, requiring an orbital-to-spin conversion layer, which motivates studies of conversion mechanisms and associated physics.

In this PhD work, we evaluate promising orbital/HM/FM material systems for SOTMRAM applications. We present a comprehensive study of Ru/HM/FeCoB and Ta/W/ FeCoB systems, where Ru and Ta act as orbital current sources, while Ta, W, and Pt serve as orbital-to-spin conversion layers. Ru is predicted to exhibit one of the largest orbital Hall angles among transition metals while maintaining low resistivity. Ta, a heavy metal with a large spin Hall effect, is predicted to exhibit an orbital Hall angle approximately one order of magnitude larger than its spin counterpart. When a heavy metal is used as a conversion layer, multiple spin-current contributions can coexist and add linearly to the total effective spin Hall conductivity, potentially enhancing the overall SOT efficiency.

We characterized key parameters relevant to SOT magnetic tunnel junctions (MTJ) devices, including saturation magnetization, effective anisotropy field, and resistivity, and we quantified damping-like (xDL) and field-like (xFL) SOT efficiencies as a function of orbital and conversion layer thickness, both in as-deposited and 300°C annealed samples. These metrics are benchmarked against reference HM/FeCoB systems to isolate the effect of the additional orbital layer. For Ru/Ta and Ru/W stacks, limited enhancement xFL of xDL is observed relative to reference systems. In contrast, Ru/Pt exhibits a twofold increase in xDL compared to Pt alone. This difference is attributed to the stronger SOC in Pt, which enables more efficient orbital-to-spin conversion. The independence of xDL on Ru thickness further suggests an interfacial origin of the orbital contribution in Ru/Pt. However, thermal annealing strongly degrades xDL, limiting its applicability for SOT-MRAM. In Ta/W systems, we observe a strong enhancement of xDL by a factor of 4.4 relative to Ta and 3.2 relative to W. A parallel-resistor model indicates that conventional SHE contributions cannot fully account for this increase, pointing to an additional orbital-related mechanism. Extending the study to 400 °C annealing shows that ξDL remains largely stable, indicating good thermal robustness while maintaining perpendicular magnetic anisotropy.

Leveraging these advantages, we further integrate the Ta/W system into SOT-MTJs and benchmark it against standard W-based MTJs. We investigate the pulse-length dependence of the critical switching current and provide a first demonstration of integrated orbital-to-spin conversion in SOT-MTJs. Ta/W devices exhibit switching currents comparable to W-based devices but have a lower switching current density and improved perpendicular magnetic anisotropy stability. Finally, we present a proof-of-concept for vertical non-local switching of SOT-MTJ using orbital torques, simplifying bottom-pinned SOT-MRAM fabrication. Overall, these results demonstrate that orbital physics can be exploited to enhance SOT-MTJ performance, simplify fabrication, and provide a promising route toward scalable bottom-pinned MRAM technologies.

resumé : Le développement de nanoaimants contrôlés électriquement pour la spintronique, en particulier les mémoires magnétiques non volatiles (MRAM), suscite un fort intérêt en raison des limites des mémoires CMOS telles que SRAM et eDRAM. Les MRAM à couplage spin–orbite (SOT) sont prometteuses ; toutefois, les matériaux actuels présentent une efficacité limitée et une résistivité élevée, empêchant de satisfaire les exigences en courant d’écriture. Des études récentes ont mis en évidence les phénomènes orbitaux comme voies potentielles d’amélioration des SOT, grâce à leur plus grande amplitude et leur disponibilité dans un plus large éventail de matériaux. Cependant, les courants orbitaux ne se couplent pas à l’aimantation en l’absence de couplage spin-orbite, ce qui nécessite une couche de conversion orbitale–spin supplémentaire et motive l’étude de ces mécanismes.

Dans ce travail de thèse, nous avons évalué des systèmes orbital/HM/FM prometteurs pour les SOT-MRAM. Nous présentons une étude des systèmes Ru/HM/FeCoB et Ta/ W/FeCoB, où Ru et Ta sont des sources de courant orbital, tandis que Ta, W et Pt servent de couches de conversion. Le Ru est prédit présenter un angle Hall orbital élevé tout en conservant une faible résistivité. Le Ta, métal lourd à fort effet Hall de spin, devrait exhiber un angle orbital environ un ordre de grandeur supérieur à sa composante spin. Lorsqu’un métal lourd est utilisé comme couche de conversion, plusieurs contributions de courant de spin peuvent coexister et s’additionner à la conductivité Hall effective totale, améliorant potentiellement l’efficacité SOT. Nous caractérisons des paramètres clés pour les dispositifs jonctions tunnel magnétiques (SOT-MTJ), notamment l’aimantation à saturation, le champ d’anisotropie effectif et la résistivité, et nous quantifions les efficacités SOT d’amortissement (xDL) et de champ (xFL), en fonction des épaisseurs des couches, pour des échantillons déposés et recuits à 300 °C. Ces résultats sont comparés à des systèmes HM/FeCoB de référence afin d’isoler l’effet de la couche orbitale.

Pour Ru/Ta et Ru/W, l’amélioration de xDL est limitée. En revanche, Ru/Pt montre un gain d’un facteur deux, attribué au fort couplage spin–orbite du Pt favorisant la conversion orbitale–spin. L’indépendance de xDL avec l’épaisseur de Ru suggère une origine interfaciale. Toutefois, le recuit thermique dégrade fortement ces performances, limitant leur applicabilité. Dans les systèmes Ta/W, xDL est fortement augmenté (×4.4 vs Ta, ×3.2 vs W). Un modèle de résistances en parallèle indique que l’effet Hall de spin ne suffit pas à expliquer ce gain, suggérant un mécanisme orbital additionnel. Après recuit à 400 °C, xDL reste stable, indiquant une bonne robustesse thermique, avec maintien de l’anisotropie perpendiculaire.

Le système Ta/W est ensuite intégré dans des SOT-MTJ et comparé à des SOT-MTJ à base de W. Les dispositifs présentent des courants de commutation comparables, mais une densité plus faible et une meilleure stabilité d’anisotropie. Enfin, nous présentons une preuve de concept de commutation non locale verticale basée sur des couples orbitaux, simplifiant la fabrication de SOT-MTJ de type «bottom pinned». Ces résultats montrent que les phénomènes orbitaux peuvent améliorer les performances des SOT-MTJ et constituent une voie prometteuse vers des MRAM scalables.

Jury :

Henri JAFFRES, DIRECTEUR DE RECHERCHE, CNRS, Rapporteur

Juan Carlos ROJAS SANCHEZ, CHARGE DE RECHERCHE HDR, CNRS, Rapporteur

Stefania PIZZINI, DIRECTRICE DE RECHERCHE, CNRS, Examinatrice

Laurent RANNO, MAITRE DE CONFERENCES, Université Grenoble Alpes, Examinateur

Young Keun KIM, FULL PROFESSOR, Korea University, Examinateur

Van Dai NGUYEN, INGENIEUR DE RECHERCHE, IMEC, Examinateur

Thesis supervisor :

Kevin Garello, CEA/SPINTEC, Directeur de thèse

The post PhD Defense – Exploration of orbital-to-spin conversion materials and integration in 3-terminal spin-orbit torque magnetic tunnel junctions appeared first on Spintec.